K-Cosine-Means Clustering Algorithm

K-means algorithm is a clustering algorithm that is one of the most widely used unsupervised techniques in data mining. This paper presents an extension of K-means algorithms named K-cosine-means algorithm. While the K-means algorithm initializes the centroids randomly and uses the Euclidean distance measure to assign data points to clusters, our proposed algorithm inherits a systematic approach from K-means++ to initialize the centroids and utilizes Cosine similarity to assign data points to clusters. We have performed experiments on both homogeneous datasets (Iris and Seeds datasets) and heterogeneous dataset (Hepatitis dataset). From experimental results, we have observed better clustering accuracy on homogeneous datasets compared to other variants of the K -means algorithm, namely, K-means, 1K-means, K-means++, WK-means, MWK-means, iWK-means, and iMWK-means. However, for heterogeneous dataset, we have observed better clustering accuracy compared to standard K-means, K-means++, and iK-means algorithms.